NAIC Committee Adopts AI Risk Evaluation Supplement, Leaving Pricing Models to Each Regulator

NAIC By InsureAI Wire. Published November 19, 2026.

Source: NAIC official document, content.naic.org (opens in a new tab)

The NAIC’s Big Data and Artificial Intelligence (H) Working Group adopted the AI Risk Evaluation Supplement at its November 16 session in Dallas. Its agenda had put version 7.0 up for adoption. The group’s meeting summary report says 7.0 was released November 9 and that the group heard more verbal comments at the session. The next item reads “Adopted the supplement after reaching a consensus to agree on the two technical edits raised.” The report does not say what those two edits were.

A day later the parent body, the Innovation, Cybersecurity, and Technology (H) Committee, adopted the supplement as well. Its summary report records the act in six words, “Adopted the AI Risk Evaluation Supplement,” without saying which version. We take the adopted text to be version 7.0 as amended by the two edits, whose content the records do not give. The supplement in the committee’s materials, compiled after the November 16 session, carries 7.0’s GLM paragraph word for word.

That paragraph sits in the General Guidance on page 3 of the clean copy prepared for adoption. It first says GLMs “are not without risk of causing unfair discrimination or other Adverse Consumer Impacts” and need effective governance. The operative sentence then rests on the review that rate filings already get: “Based on existing regulatory oversight provided via rate making,” it says, “regulators may exclude some or all GLMs and other predictive models” from the supplement “if appropriate.” So the choice is the regulator’s, and it can reach some GLMs or all of them; the sentence neither takes pricing models out by itself nor obliges a regulator to keep them in.

Two trade associations had asked for more in their letters on version 6.0. NAMIC’s letter “strongly suggests that the Supplement clarify that GLMs and other predictive models used in Ratemaking are explicitly out of scope.” APCIA wrote that 6.0 let regulators exclude GLMs but “that decision remains discretionary,” and it recommended excluding them from Exhibits B, C and D. In 6.0 the exclusion reached only those three exhibits; 7.0 extends it to the whole supplement, and margin notes in the marked-up copy of 7.0 credit input from Verisk, ACLI, APCIA, AHIP and NAMIC. The wider reach moves toward the scope of NAMIC’s request. The exclusion is still the regulator’s choice, though, which falls short of what both letters asked.

The minutes of the working group’s November 2 meeting, which the group adopted on November 16, record, in remarks by its chair, Commissioner Nathan Houdek of Wisconsin, that the version it would adopt in Dallas “will be available for voluntary state use next year.” Those remarks also leave for next year whether the supplement goes into the Financial Condition Examiners Handbook or the Market Regulation Handbook. Our November 13 report covered those minutes while they were still a draft.

On our reading, a pricing or actuarial team that wants a regulator to exclude its GLMs from this supplement can cite the text’s own ground for allowing it, existing oversight through rate making. The same page says GLMs need effective governance, so governance records for those models are worth having ready whichever way the regulator decides.

Status as of November 18, 2026. The working group and the committee have both adopted the supplement. The posted records give no effective date.

Share

Go deeper

Keep tracking

Information aggregation and analysis, not legal advice. See our disclaimer.